Soil depth and parameters #17
Comments
|
Added shapefile statsgossurgosoil_dhsvmparams.shp to Hydroshare Added C:\Users\cband\Skagit\SCLlandsliding\SkagitLandslideHazards\ |
Soil depth AML for generating DHSVM soil depth - we used range of [1,3] meters for soil depth grid./* soildepth.aml Kenneth Westrick 12/27/1999 |
|
CB and RS reviewed the soil surveys for the Skagit River Basin.
These surveys are driving the patterns in water table depth via Ksat and f() |
|
oops, didn't mean to close this issue |
|
C:\Users\cband\Skagit\SCLlandsliding\SkagitLandslideHazards\DraftResults_SkagitLandslide_SoilsWTD.mxd
Why is it not saturated in the same pattern in the Okanogan County? Here is the Ksat parameter that goes into DHSVM and controls saturation during the peak events - this map above is from 2042-6-7-15 the fourth date in the list for CNRM rcp4.5
|
The USDA input file plots the polygons inside the triangle for the different texture classes.
|
|
Before talking to Dan - I searched and found this tool that calculates a texture raster from a sand and clay grid: https://github.com/gmassei/SoilTexture Should we try to run this? I don't want to deal with the interface... |
|
It looks like Rosetta also has an improved version that we could try http://www.u.arizona.edu/~ygzhang/rosettav3/ More about this version here: https://soil-modeling.org/news/news-images/zhang-2017-ismc-agu.pdf |
|
So, I took a brief look at this and looks like we might be able to either of these methods to convert %sands%clay to soil texture. Unless there is a new tool/method/report I haven't been able to locate, there is another step after texture. We need to take that texture and match to USCS classification. From the USCS class we can use tables to estimate the mean fiction angle. One of the best tables for getting to internal angle of friction is here: |
|
I heart soil guesswork. SSURGO does have a lot of the organic matter, moisture, compaction data...I think we need to clone a repo that does most of what we want, and then add on the rest of these steps and make a 'soil processing for landslides' set of Python scripts. Does anyone have strong feelings about the SoilTexture code versus the Rosetta code? especially for the needs as @RondaStrauch has outlined them? |
|
Proposed simplified soil triangle needed for USC would be
|
|
In the Skagit there is NO
Could have |
|
@DanMillerM2 What Phi approaches do you approve of? |
|
Thoughts from @DanMillerM2 email 7/11/19 I'm interested in seeing what the model predicts before spending too much time working on the paper. You are considering two primary factors:
Using estimated distributions of soil depth and geotechnical properties, can you make a map showing the saturation depth required to trigger landsliding for current landcover and for minimum cohesion following fire? Then what happens under predicted future climate, both for current landcover and minimum cohesion? How well is current fire regime characterized? I've found a few studies in my bibliography examining soil strength as a function of soil properties (grain-size distribution, bulk density, liquid limit). I don't think they are particularly useful, other than to show general trends. Soils are very complicated and measured peak and residual strengths depend on the stress trajectory; I have a lot to learn, but I can keep looking for guidance. However, I'm interested in how soil properties affect the modeled frequency distributions. How important is hydraulic conductivity and transmissivity relative to friction angle and cohesion? 7/12/19 The sensitivity of the infinite slope equation to perturbations of any single parameter (or sets of parameters) is easy to characterize, but you've now set up a system that is sensitive to upslope topography, upslope soil parameters (depth, porosity, conductivity), upslope vegetation cover, and antecedent weather (if you're using DHSM to estimate soil moisture). And we're interested not just in how probability of failure might change at a single location at a single time, but how the integrated probability over some area and some period of time might change. Given all these potential interactions, the relative importance of each parameter and factor that influences model results might be hard to anticipate. So one approach is to use multiple model runs with one factor changed and see how the results change. And those results are characterized as frequency distributions of probability of failure over some area and some period. For example, a change in the frequency of storms may alter the spatial distribution of predicted failures. How important are soil parameters (friction angle, porosity, conductivity, transmissivity) to the predicted change in spatial distribution? For the most part, we model these as linear systems, so we don't expect any radical responses, but we should check that our intuition is correct. Such analyses are probably beyond the current project scope, but it would be good to set up a modeling platform that allows such analyses. The textbook "An Introduction to Geotechnical Engineering" by Holtz and Kovacs provides a nice description of factors influencing frictional resistance of non-cohesive soils, which is paraphrased in the Slope Stability Reference Guide for National Forests, volume 2. And I think you've already used the table of values provided in the Lisa model documentation. For a given soil type at low confining pressure (shallow depths), friction angle varies with bulk density (porosity, or void ratio). If you can get estimates of soil type and bulk density (or porosity, or void ratio) from SSURGO, then we could specify a range of friction-angle values. Duncan, Wright, and Brandon (Soil Strength and Slope Stability) give an equation for friction angle as a function of relative density, confining pressure, and grain-size distribution (given as the coefficient of uniformity: the ratio of D60 to D10). Not sure how useful that would be, unless you can get estimates of relative density and coefficient of uniformity from information included in the soils database. |
|
Response to Dan I believe DHSVM accounts for upslope characteristics when it is estimating a depth to groundwater at each 150m grid cell. We used the same estimated depth distribution for each 30 m DEM cell within the 150 m larger cell. I'm not sure how we can integrate the probability of failure at each grid cell into a larger area of probability and why would we want to aggregate this up to a larger spatial scale, but I might be missing your point. We could do the multiple Monte Carlo runs by reducing the depth to groundwater distribution, say by 10%, to see how the probability of failure changes spatially. This would give an indication of just how conservative this is. We're not changing the frequency of storms, but seeing if a less intense storm also results in substantial hazard. I think our model can accommodate a parameter sensitivity analysis. Part of our approach is to reduce the root cohesion to simulate a post fire landscape and see how the probability of failure changes. Regarding soil parameters, we can check if SSURGO has soil type and bulk density to see if we can estimate the range of friction angles from the references you provide. Regarding the spread sheet, yes, SSURGO is the source for the percent sand, silt, and clay as well as soil texture. We use the texture to estimate USC like SW and compared that to the tables 5.4, 5.5 in LISA. We took a mean angle for these classifications. |
|
@RondaStrauch @DanMillerM2 The table I have output already has the soil texture, but the percent components seem to be more detailed information for our angle parameter estimate. I can get bulk density also, we already have this output from the database as it is a DHSVM input parameter. |
|
@DanMillerM2 We worked on the soil friction angle today to look how it displays spatially. For our study area, the friction angles are spatially similar but different for regions of the North Cascades. If you turn on the texture label, you can see that the north eastern portion tends to be gravels and sands, while the tan sections tend to be loams. (zoomed in a bit) Thoughts? |
|
Interesting. Let's compare to geology and topography. Variations in soil
texture should correspond to variations in substrate and with topographic
position. Upper hillslopes have mostly in-situ soils; lower slopes have
colluvium that has moved (in some cases by landsliding) from upslope;
valley floors have colluvium (probably landslide deposits mostly) and
alluvium; low-lying depressions may have wetlands and fine-grained soils.
Do the soil types and textures make sense relative to the mapped geology
and DEM-derived topography? For landslide initiation, we're mostly
interested in the upper hillslopes. How much variation in texture is
indicated for the upper hillslopes?
Do you have a shaded relief image of the area?
Can I just download the datasets? Like the DEM, the soil-type polygons
(with attributes) you show in the figures?
How much do you expect soil depths to vary on the upper hillslopes? I
recall use of a curvature-based model for estimating relative soil depths,
like what Bill Dietrich and his students came up with.
How much does SSURGO indicate that texture varies with depth for a single
soil type? Do we expect variations in texture, and friction angle, at the
soil-rock interface with variations in soil depth on those upper
hillslopes? Landsliding is probably generally focused in topographic
convergent zones, which the soil-creep models indicate will have deeper
soils. Of course, those are also the zones more likely to get scoured by
landslides, so soil depth will depend on frequency of landsliding.
Now, these are just random thoughts triggered by the figures you sent. We
can probably come up with an actual strategy when we focus on specific
questions.
…On Wed, Jul 24, 2019 at 4:00 PM Christina Bandaragoda < ***@***.***> wrote:
@DanMillerM2 <https://github.com/DanMillerM2> We worked on the soil
friction angle today to look how it displays spatially. For our study area,
the friction angles are spatially similar but different for regions of the
North Cascades.
[image: image]
<https://user-images.githubusercontent.com/4108369/61833877-9d27c700-ae2a-11e9-94a4-a909883507c0.png>
[image: image]
<https://user-images.githubusercontent.com/4108369/61833926-ca747500-ae2a-11e9-865a-dd483746dc0b.png>
If you turn on the texture label, you can see that the north eastern
portion tends to be gravels and sands, while the tan sections tend to be
loams.
[image: image]
<https://user-images.githubusercontent.com/4108369/61834245-ff34fc00-ae2b-11e9-871e-05b150584526.png>
Thoughts?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#17?email_source=notifications&email_token=AML2SYHK3DJIIUDZBHLMUE3QBDNIJA5CNFSM4HD7FXH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD2X265Y#issuecomment-514830199>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AML2SYGRZ27RD55HXIHWH73QBDNIJANCNFSM4HD7FXHQ>
.
|
|
@DanMillerM2 All good questions, as always. I will export the shapefiles, put them on HydroShare, and you can download them, as well as soil depth and DEM inputs we are planning to use for the Landlab Landslide component. Do you prefer ASCII or shapefiles? We need ASCII as inputs, so maybe I'll just process both and put it all on HydroShare. Other details I should think of while I'm managing the data? |
|
Either data type is fine.
For large data files, ascii and shapefiles are cumbersome, but perhaps
these are not so big.
…On Thu, Jul 25, 2019 at 12:42 PM Christina Bandaragoda < ***@***.***> wrote:
@DanMillerM2 <https://github.com/DanMillerM2> All good questions, as
always. I will export the shapefiles, put them on HydroShare, and you can
download them, as well as soil depth and DEM inputs we are planning to use
for the Landlab Landslide component. Do you prefer ASCII or shapefiles? We
need ASCII as inputs, so maybe I'll just process both and put it all on
HydroShare. Other details I should think of while I'm managing the data?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#17?email_source=notifications&email_token=AML2SYC7OLD63JGODGULOC3QBH6ZRA5CNFSM4HD7FXH2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD22R6BY#issuecomment-515186439>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AML2SYGZMTV4BIDNQWXPTM3QBH6ZRANCNFSM4HD7FXHQ>
.
|
@RondaStrauch add the picture here please.
The text was updated successfully, but these errors were encountered: